Raw audio
Never stored — audio is transcribed as it streams and is not written to disk on our side.
Thalassa is built for conversations you can't afford to leak: client calls, advisory sessions, negotiations. The architecture is simple — keep as little as possible, keep it in Europe, and keep your documents on your machine.
Never stored — audio is transcribed as it streams and is not written to disk on our side.
Deleted after the meeting. Our database has no tables for meeting content — enforced by automated tests.
Screen analysis happens for your answer only; captures are not retained.
Your knowledge base lives on your device. We hold no copy — we couldn't read your files if we wanted to.
Answer generation and account infrastructure run in the EU (Frankfurt). We designed for European data-protection expectations from day one: data minimisation, no permanent meeting records, and your knowledge base kept out of our hands entirely.
Your answer overlay is private to you and stays off the screen you share — so you stay present and your shared screen stays clean. It floats above your call window, only on your own monitor; move it, resize it, forget it's there.
Illustration of the actual overlay behaviour.
You are responsible for obtaining any consent required to transcribe a conversation (in Germany, all participants must consent) and for disclosing AI assistance where required. Thalassa gives you the controls; you decide how to use them.
Full legal details: Datenschutzerklärung / Privacy Policy